Products: Speech Analytics - SERTAINTY - FAQs
- What is SERTAINTY?
-
SERTAINTY combines customer interaction data collection, phonetic audio search technology, and a powerful rules engine to enable cost effective, automated, enhanced quality monitoring for the contact center. SERTAINTY enables contact centers to listen to 100% of their recorded calls without adding additional human resources. This product forever changes the paradigm of monitoring customer and agent interactions by eliminating the need to listen to calls that are not interesting or important to the center. Prior to the introduction of SERTAINTY, listening to recorded calls was a manual process that required quality assurance agents to perform the tedious, time consuming, and laborious task of listening to live or recorded calls and subjectively evaluating them. This manual process is extremely resource intensive, even when the quality assurance agents listen to calls using high-speed playback.
- Is SERTAINTY a call recording product?
-
No. SERTAINTY is a speech analytics product that analyzes previously recorded calls.
SERTAINTY does not capture or store audio. It requires an audio source, acts as a listening agent, delivers actionable events, and provides a vehicle for analysis of every recorded call. The product is designed to easily integrate with call recording products and leverages existing investments in call capture and archiving.
For our customers who currently do not have a recording system, it’s not uncommon for contact center solutions to come from separate vendors, have separate interfaces, separate data, and unfortunately, no link between them. Wygant and SER understand the shortcomings of non-integrated solutions and have partnered to deliver seamless integration between Wygant’s Encore™ call recording solution, SER’s CPS E² outbound call management system, and speech analytics solution, SERTAINTY™.
- When can analysis occur?
Analysis occurs as soon as the audio is available to SERTAINTY. Currently, when SERTAINTY is deployed in conjunction with a call recording system, analysis occurs when the audio file can be extracted from the call recorder. This is normally soon after the end of the call. So, call analysis happens in near real-time.
- How does SERTAINTY take subjectivity out of the quality analysis?
As you know, human analysis will always include a certain amount of subjectivity. Individuals often see or hear the same thing at the same time but report two different results. Through the implementation of objective rules, SERTAINTY removes any doubt of subjectivity by providing results based on the actual content of the dialogue and the set of business rules executed against the dialogue.
- How does SERTAINTY make the quality assurance process more cost effective?
Without the innovation of SERTAINTY, there is no cost effective way to evaluate 100% of a contact center's calls when using quality assurance agents to perform the evaluation, nor can you filter contacts so that quality assurance agents listen to the most important calls, the ones that have dialogue of interest.
Until now, even where 100% of all customer interactions are recorded, companies have to compromise and only evaluate a random sampling of calls, as little as 1% or less. However, studies suggest 85% or more of the calls that your agents handle, fall within the norm and therefore have no impact on how you coach, reward, or motivate your agents. That means that, using random sampling, the odds of listening to a call that falls within the 15% of "high value" calls are very low.
With SERTAINTY, you are able to review 100% of the calls and identify calls that have interest for your company, whether good or improper. Based on your business requirements, quality assurance agents are able to concentrate their efforts on the calls of interest identified by SERTAINTY, making the odds of catching the "important" call very good.
- What makes a call a "high value" call?
The set of rules that SERTAINTY uses to score a call are easy to implement and user configurable. In general, there are two types of rules in SERTAINTY, required phrases and forbidden phrases. Most business units consider a call improper if it contains one or more forbidden phrases or is missing one or more required phrases. Calls can be scored based on the business unit’s criteria and based on mean scoring, by setting benchmarks and accumulating statistics with SERTAINTY. Calls can be considered superior when they exceed the mean or if an agent consistently scores above the mean. When those calls are identified, you may want to use them in coaching sessions to help demonstrate superior service.
- How do I define rules?
-
SERTAINTY includes a rules wizard to guide you through the creation of quality profiles, KPIs, and phrases. The rules wizard presents a simplified interface that allows non-programmers to define a powerful set of rules. By using audio from your own environment, the rules wizard lets you test the rules and tune them as needed.
In addition to the rules that are defined through the wizard, power users can define additional rules using Visual Basic for Applications (VBA). The off-the-shelf integration between the SERTAINTY Rules Manager and VBA provides an easy way for customers to extend the power of SERTAINTY without the need for product enhancements.
- Does each rule rely on just a single phrase?
SERTAINTY can search for keywords and phrases that are forbidden. It can search for phrases that are encouraged or required. And, it can also search for words and phrases that you might define for any other marketing, quality control, compliance, or general interest reasons.
Rules are not associated with a single phrase, but with a Key Performance Indicator (KPI). A KPI is defined by a set of phrases and the rule is satisfied if any of the associated phrases are detected. For example, the KPI might be "agent obtained permission to record" and the associated phrases could be "May I record this call" or "This call is being recorded" or "Permission to record this call". The rules author gets to specify the set of phrases, any one of which will satisfy a KPI and therefore a rule.
- Is a KPI definition just a collection of simple phrases?
SERTAINTY supports a complex grammar for phrase specification. A phrase can be a single phrase anywhere in the audio, a single phrase within a certain proximity to either the start or end of the audio, or one or more phrases within certain proximity of each other. Examples would be "Never call me again" anywhere in the text, "Thank you for calling Joe's Bank" within the last 10 seconds of the call, or "Joe's Bank, my name is" followed by "How may I help you" within 5 seconds of each other.
- There could be a call where all the required phrases are said but there is some other problem. Can SERTAINTY flag those types of calls?
Our experience suggests that most contact centers and the calls they make or take follow a scripted flow or can have a random flow but within the context of certain questions will usually have some structure. Typically, even inbound calls, taken for any number of reasons, usually require that the agent take control and direct the dialogue. SERTAINTY continually monitors calls and gathers information about the dialogue flow through rules execution. Therefore, SERTAINTY can detect a variation from the normal flow of the call and provide an alert if it detects there may be something of interest to listen to, even if the rule wasn't written to listen for it. In other words, when the script for a call suggests the dialogue and therefore the KPIs follow a specific order, and the time between any KPI does not vary much if it follows the average flow, SERTAINTY can detect when the call falls outside the norm and flag the call for review.
- Does SERTAINTY rely only on the audio to evaluate a call?
Each call normally has some set of meta data associated with it. This data normally includes the agent ID, the DNIS, and/or the ANI. It may also include a customer ID or an IVR response option. SERTAINTY uses this information to determine which quality profile or profiles to use in evaluating the call. A quality profile is a set of KPIs. In some installations, there is additional meta data such as the call disposition and other information about the interaction. This can be used to enable or disable specific rules. For example, if the meta data tells us a call resulted in a credit card sale, then we might execute a KPI associated with the agent asking for credit card information. But that same KPI would be disabled if the call was not a sale or was a sale that did not include a credit card.
- Today's business world has a lingo all of its own. Can SERTAINTY find manufactured words?
-
Yes. Unlike other speech recognition technologies, SERTAINTY stands alone in today's market because its phonetic audio searching capability does not rely upon closed vocabulary dictionaries, static grammars, or the need to convert speech to text. As a result, recorded conversations can be subsequently audited or mined for any combination of words or phrases, or the absence of words or phrases. For instance, nearly every speech dictionary will include the words Robert Jones and a grammar can be written to find them. However, it's unlikely that you would find the words, trebor senoj, which happens to be Robert Jones backwards.
SERTAINTY will find them both.
- How does SERTAINTY handle thick or heavy accents?
SERTAINTY uses phonetic searching to find phrases. By default, a search phrase is automatically translated into a phonetic search stream by the product. This translation is done using a built in dictionary for common words. Words not found in the dictionary are processed by consulting a spelling to sound database that extracts likely phonetic representations given a word's orthography. However, the product allows the user to represent a search phrase using a phoneme based representation. This allows search phrases to be specified as they would sound when spoken with a heavy accent.
- How do I review calls?
-
SERTAINTY's Call Analysis Dashboard is the interface quality assurance agents use to review calls. The default view shows calls that have alerts and have not yet been reviewed. Search capabilities allow the reviewer to customize the display of calls. The reviewer can select a call, see the alerts, listen to the call and disposition it. Clicking on an alert in the Call Analysis interface takes the reviewer to the exact location in the call requiring attention.
With Call Analysis, you can modify and display reports in a variety of graphs and table formats. And, Call Analysis screens are flexible and customizable. You can add and delete tabs and reports and arrange multiple reports on a single screen. Its flexibility within the reports allows you to drill down into a graph or table to get specific sets of calls or KPIs quickly and efficiently.
Call Analysis includes many features for call/KPI reviewing. You can review, reopen, and escalate calls/KPIs, and you can configure users for specific review permissions. You can review calls with the multifunctional built-in audio feature, either at the point of alert or from the beginning of the call. Options to modify the state of calls/KPIs include the ability to re-grade calls from Good, Bad, or Warning. For publishing purposes, you can download and upload reports to and from Crystal Reports.
Charts can give a good overall view at a glance of the percentage of good and bad calls/KPIs in a group. For a closer look, in the Call Registry and View screens, you will find the individual details of each call, including Call ID, Agent ID, Alert, Severity, Grade, Start and End Times, and Audio.
- Can I go back and search the audio for different phrases at a later time?
SERTAINTY provides the unique ability to "process once" and "search many times" across an asset of available audio. When the initial call analysis is completed, SERTAINTY creates an intermediary file format that can be searched at very high speeds - enabling a user to search across any set of processed files for any phrase. This turns your audio files into a searchable data archive, which can be mined for valuable business intelligence.
SERTAINTY’s “Audio Mining” feature provides the ability to search calls based on the actual content of the call in addition to the common search methods prevalent in the industry, such as the time the call started, the duration of the call, or the channel that originated or received the call. This unique ability makes it easy to find problems in calls and to quickly generate lists of calls that need special attention.
- Is there a limit to how much processed audio I can store and analyze?
No. The only limitation is available storage.